Taobao's deployed RecGPT-V3 uses stateful behavior modeling, a hybrid-modal base model (integrating text tags and Semantic IDs), and latent intent reasonin...
Taobao's deployed RecGPT-V3 uses stateful behavior modeling, a hybrid-modal base model (integrating
text tags and Semantic IDs), and latent intent reasoning, reducing user modeling computation by 55.8% and output token cost. In large-scale online A/B testing of the 'You May Like' feed, IPV increased by 1.28%, CTR increased by 1.00%, TC increased by 1.97%, GMV increased by 3.97%, and end-to-end service resource consumption decreased by 52.4%.
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs ：https://info.arxiv.org/labs/index.html.
RecGPT-V3 已在淘宝部署，通过状态化行为建模、融合文本标签与 Semantic IDs 的混合模态基础模型及潜在意图推理，降低用户建模计算量与输出 token 成本。摘要称，其在线 A/B 测试中 IPV、CTR、TC、GMV 分别提升 1.28%、1.00%、1.97%、3.97%。
该材料来自 HuggingFace Daily Papers 的社区热门论文，分类为论文研究。标题将其描述为状态化混合模态推荐系统；摘要说明测试场景为淘宝“猜你喜欢”信息流，但正文摘录仅介绍 arXivLabs，未补充论文方法或实验细节。
Aioga 判断，这项工作的关注点不仅是推荐指标变化，还包括推理成本控制：摘要同时报告用户建模计算量降低 55.8%、端到端服务资源消耗降低 52.4%。不过，现有材料不足以核验实验周期、样本范围及统计显著性。
Aioga 判断，若论文完整结果与摘要一致，状态化建模与混合模态表示可能为大型推荐系统提供效果和资源消耗并行优化的参考。值得关注的是，标题所称“GPU 成本降低”与摘要中的“服务资源消耗降低”表述并不完全相同。 下一步应查阅论文全文，重点核对 IPV、CTR、TC 与 GMV 的定义，在线 A/B 测试设计、实验周期、流量范围和显著性结果，以及 52.4% 降幅对应的资源口径，从而判断这些结果的适用边界与可复现程度。
The readable text on this page was extracted from the public source and organized with attribution, publication time and the original link. Copyright remains with the original author and publisher.
Ingestion channel: Summary aggregation · Source domain: arxiv.org
Source: HuggingFace Daily Papers（社区热门论文）
Original link: Open original source
Aioga archive: Open intelligence page
Content record: source-page · Updated: 2026-07-17T00:00:00.000Z
统一接入主流 AI 模型 API，为开发、测试与生产环境提供稳定调用入口。
立即访问 api.w173.com
Scan to open this articleAioga aggregates global AI updates and preserves source information for verification and citation.